The Embryology of Dacus tryoni (Frogg.) [Diptera, Trypetidae (= Tephritidae)], the Queensland Fruit-Fly

The embryology of the Diptera has been investigated in only a few species and of these, only Drosophila melanogaster and Calliphora erythrocephala have been studied in detail. Fragmentary information is available for 17 species of Nematocera, several further species of acalyptrate Cyclorrhapha (all Drosophila species), 6 further species of calyptrate Cyclorrhapha, and 2 species of pupiparous Cyclorrhapha, but little attempt has been made to compare one with another. The present description of the embryology of the larva of the acalyptrate Dacus tryoni, the first of a member of the Tephritidae, provides new data on which a comparative survey of the embryology of dipteran larvae can be based.

Material for this study was supplied by Dr. M. A. Bateman of the Joint Unit of Animal Ecology in the Department of Zoology, University of Sydney, from his laboratory stocks. Rearing of D. tryoni in the laboratory has been fully described by Bateman (1958) and only a brief outline will be presented here. Adult flies were kept in cages at 25° C., 80 per cent, relative humidity. When domes of apple skin, sealed by paraffin wax to flat glass plates and pierced by a number of fine holes, were placed in the cages, females settled on them and laid eggs through the holes into the domes. An oviposition time of 15 minutes was allowed for each dome, so that the age of eggs subsequently examined was known to within ± minutes. After completion of oviposition, the eggs were transferred by means of a fine brush wetted with distilled water on to squares of damp filter paper in a Petri dish and left at 25° C., 80 per cent. r.h. until required. In these conditions, hatching of the larva, a typical cyclorrhaphan maggot, took place approximately 42 hours after oviposition. The few hatched larvae used to complete the final stages of the study were allowed to burrow and feed in carrot medium at the same temperature until required.

From eggs obtained in this way, development was followed from oviposition to hatching in living embryos and by serial sections. Living embryos were examined after removal of the opaque chorion of the egg by immersion for 30 seconds in 10 per cent, sodium hypochlorite followed by transfer to distilled water. The transparent vitelline membrane within the chorion remained intact, protecting the embryo and at the same time permitting observation of changes in external form.

In the preparation of serial sections, other eggs were fixed in aqueous Bouin at boiling-point for 10 minutes, transferred to fresh cold Bouin for 24 hours, then stored in 70 per cent, alcohol. Fixation in this way facilitated penetration of the egg membranes by the fixative and also caused the membranes to swell and become transparent, so that they could easily be dissected off. After passage through 95 per cent, alcohol, methyl benzoate and benzene, the embryos were embedded in paraffin (m.p. 56° C.), sectioned at 8 μ transversely, sagittally, or frontally and stained with Mayer’s haemalum and eosin.

The egg of D. tryoni is elongate, curved, 900–1,150 μ long and 170–210 μ in diameter. One end, the anterior, is drawn out as a pointed papilla, the opposite end is gently rounded. The antero-posterior axis of the egg is established in the ovary and the posterior end emerges first at oviposition. The dorso-ventral axis is also fixed during oogenesis and reflected in the curvature of the egg, its convex surface being ventral.

The egg is covered by a tough, pearly-white chorion marked externally by a regular lattice of ridges. Immediately within the chorion lies a thin vitelline membrane and within this again the whitish granular living substance of the oocyte. Both chorion and vitelline membrane have a micropyle at the anterior pole of the egg (Text-fig. 2B). The oocyte has the characteristic dipteran structure (Text-figs. 1A; 2A), with a thin surface layer of cytoplasm, the periplasm, slightly thickened at the anterior and posterior poles, enclosing a central mass of cytoplasm containing abundant yolk spheres of various sizes interspersed with numerous vacuoles. Within the posterior thickening of the periplasm lies an additional mass of small polar granules having a high affinity for haematoxylin stains. The cytoplasmic reticulum which forms the walls of the vacuoles and holds the yolk is continuous with the periplasm at the surface. The vacuoles have a constant distribution and appearance in fixed eggs, being relatively small and scattered throughout most of the interior, but including a peripheral layer of larger vacuoles absent only at the anterior and posterior ends. Although this appearance must to some extent be an artefact, its constancy indicates an underlying special distribution of certain, as yet unknown, egg constituents.

Close to the anterior pole of the egg lies the oocyte nucleus, surrounded by a halo of yolk-free cytoplasm (Text-fig. 1A). With a diameter of 30 and small scattered chromatin granules as its only discernible content in fixed preparations, it is clearly distinguishable from the zygote nucleus resulting from fertilization.

Just before oviposition, a single sperm enters the micropyle, penetrates the surface of the oocyte antero-ventrally, and passes into the central region of the anterior end of the egg, pulling in with itself part of the surface periplasm as a sperm track (Text-fig. 2B). Some of this cytoplasm no doubt contributes later to the cytoplasmic halo of the zygote nucleus.

Maturation of the oocyte is associated with insemination, but the time relations of the two processes and the details of maturation are not yet clear. Immediately after oviposition, the halo of the oocyte nucleus, now containing a number of small scattered chromosomes, lies at the dorsal surface of the egg near the anterior pole. Within the next 20 minutes a spindle forms and a typical maturation division proceeds (Text-fig. 2B). The resulting polar nucleus, embedded in the periplasm of the egg, soon disappears. The female pronucleus unites with the male pronucleus and the resulting zygote nucleus, surrounded by a cytoplasmic halo formed by fusion of the haloes of the pronuclei, then migrates to the midline of the egg a short distance from the anterior pole (Plate 1, fig. A). In contrast to the oocyte nucleus, the zygote nucleus, formed within 30 minutes of oviposition, is 5μ in diameter and has no distinct chromatin granules, although it shows a weak Feulgen reaction.

In the absence of further evidence it seems likely that the maturation division which follows oviposition is the second and that the first division is completed before the egg is laid. If this is so, the fate of the first polar nucleus has yet to be elucidated.

Within 30 minutes of its formation the zygote nucleus divides by mitosis and its products divide again in synchrony (Text-fig. 2c; Plate 1, fig. B) to give 4 cleavage nuclei, each with its own cytoplasmic halo, lying at the border of the anterior third of the embryo.

Changes in the egg cytoplasm accompany these divisions. Vacuolation is more noticeable, commensurate perhaps with the manufacture of additional cytoplasmic haloes at the expense of the cytoplasmic reticulum. An antero-posterior elongation of the vacuoles of the posterior third of the egg also takes place, indicating a transfer of special substances either towards the posterior pole or from the posterior pole forwards (Plate 1, fig. B). Such tendencies become more plain during the subsequent cleavages giving 8, 16, 32, and 64 nuclei. These, the 3rd to 6th mitoses of cleavage, with their accompanying divisions of cytoplasmic haloes and interim periods of nuclear interphase and cytoplasmic accretion, are also synchronous and occupy the 2nd hour after oviposition. At the same time the nuclei spread posteriorly through the egg (Text-fig. 1B; Plate 1, fig. C) but do not invade the periplasm. The rate of division during the 2nd hour, one mitotic wave every 15 minutes, is identical with that of the first two cleavages.

The 3rd hour of development brings a change in nuclear behaviour, following first the synchronous division of 64 into 128 nuclei (7th mitosis of cleavage).

Many nuclei now migrate to the egg surface and enter the periplasm, their cytoplasmic haloes fusing with the latter, while the remainder hold their position in the yolk as vitellophages. The distinction between surface and vitellophage nuclei thus established is permanent, since no vitellophages subsequently move to the surface and no surface nuclei, with one exception (see below), re-enter the yolk as vitellophages. Approximately 90 surface nuclei and 38 vitellophage nuclei are separated out, the numbers varying slightly from one egg to another. Nuclear invasion of the surface is simultaneous throughout the entire periplasm. At this stage, when the cytoplasmic haloes are being drawn into the periplasm and considerable reorganization of the embryo is taking place, the yolky inner mass of the egg becomes difficult to fix and generally shows large vacuoles, no doubt fixation artefacts, indicative of an increased fluidity.

By the end of the 3rd hour the nuclei have undergone three further synchronous mitoses (Text-fig. 3B) giving 180, 360, then 720 nuclei at the surface and 75, 150, then 300 vitellophage nuclei scattered throughout the yolk with their cytoplasmic haloes united by the cytoplasmic reticulum both to each other and to the surface layer (Text-fig. 3A). Division of the vitellophage nuclei appears to cease in many eggs at this stage, although in a few cases a further wave of mitosis coinciding with the penultimate mitotic wave at the surface takes place and the final number of vitellophages becomes approximately 600. The rate of mitotic division in the 3rd hour of development is sustained at one per 15 minutes.

Towards the end of the 3rd hour, 4 posterior surface nuclei associated with the polar granules come to lie in cytoplasmic protrusions at the posterior end of the egg (Text-fig. 3A). During the next hour of development the protrusions are cut off as 4 pole cells which have the polar granules concentrated wholly within their cytoplasm.

In the 4th hour of development, two further synchronous waves of mitotis at the surface, the 11th and 12th of cleavage (Text-fig. 3c) (rate of division halved), produce approximately 1,440 then 2,880 surface nuclei. The pole cells, however, do not divide in synchrony with the surface nuclei. During the 4th hour they divide once, becoming 8, and during the 5th hour once again, becoming 16 (Text-fig. 4A). The cytoplasm of each cell contains polar granules. The pole-cell group as a whole is separated by a small space from the underlying surface of the embryo, which in this vicinity is relatively free of nuclei, save for a few, between 2 and 5, within an inwardly projecting cone of surface cytoplasm (Text-fig. 4A). One or two of these nuclei separate off about this time, each with a cytoplasmic halo, as secondary vitellophages. The others remain within the cytoplasmic cone (Text-fig. 4 B, C).

The nuclei at the surface of the embryo undergo a final wave of mitosis, the 13th of cleavage, during the 5th hour of development, establishing the definitive number of approximately 5,800, tightly packed within the periplasm (Text-figs. 3D; 4A). The latter is also greatly thickened, the major part of the cytoplasmic reticulum having been drawn into it, leaving the yolk as a central mass of tightly packed granules with vitellophages scattered among them. The surface layer can now be termed a syncytial blastoderm.

During the 6th hour, the blastoderm nuclei enlarge (Text-figs. 3E; 4B; Plate 1, fig. D) becoming oval, with the long axis radially aligned. At the same time the boundary between blastoderm and yolk becomes regular and distinct, evidently as the incorporation of cytoplasm from the interior into the blastoderm is completed. The vitellophages all migrate towards the longitudinal midline of the yolk, while the pole cells divide once more to reach their definitive number of 32 (Text-fig. 3G).

The 7th hour of development is marked by the onset of processes transforming the syncytial into a cellular blastoderm. The surface convexities already associated with the blastoderm nuclei indicate the limits of the future cells and during the 7th hour radial cell membranes begin to push down between and beyond the nuclei, which at the same time develop nucleoli (compare Text-figs. 3 E, F). In embryos 7 hours old the membranes have reached deep into the blastoderm (Text-fig. 3F; Plate 1, fig. E) but completion of the radial and inner tangential membranes cutting off the cellular blastoderm at the surface of the yolk takes a further 30 minutes (Text-figs. 3G; 7A; Plate 2, fig. A).

The inner tangential membranes of the blastoderm cells lie not at the boundary between blastoderm and yolk but slightly peripheral to it. As a result, a thin continuous layer of nucleus-free cytoplasm remains as a sac around the yolk and vitellophages (Text-fig. 3G). At the posterior pole of the embryo the inner tangential boundary of the blastoderm meets the space beneath the pole cells, which thus lie within a circular polar aperture in the blastoderm. The nucleated cytoplasmic cone beneath the pole cells is incorporated into the perilecithal cytoplasmic sac (Text-fig. 4C).

As soon as the cellular blastoderm is complete a series of migrations and divisions commences among its cells, and those which will give rise to the internal organs of the larva move into the interior. At the same time various parts of the embryo acquire characteristics which allow them to be histologically identified as organ primordia. This complex of processes can be placed under the heading of gastrulation. Its description is facilitated by preliminary reference to a presumptive area map of the blastoderm, derived not from differences detectable in the blastoderm by the methods employed in this study but by extrapolation of subsequent events back on to the blastoderm areas in which they are initiated (Text-fig. 5). It remains for detailed experimentation to decide how far these areas can be separated in the blastoderm or earlier in terms of ooplasmic segregation or to what extent their establishment depends on causal processes of epigenesis during cleavage and gastrulation, although the evidence available for other cyclorrhaphous Diptera suggests that the presumptive areas are determined at a very early stage in development, perhaps even before fertilization. Acknowledging this limitation, however, the apparent presumptive areas shown in Text-fig. 5 are of considerable use in the present task of describing the cell lineage of the species.

The main areas which can be discerned are :

1. Mid-ventrally, as a band some 16 cells wide along the length of the blastoderm, stopping short of either end—the presumptive mesoderm.

2. Immediately lateral to the presumptive mesoderm on either side, a narrow marginal strip which contributes both to the mesoderm and to the ventral nerve-cord.

3. Antero-ventrally in front of the anterior end of the presumptive mesoderm —the presumptive anterior portion of the larval mid-gut.

4. Surrounding the presumptive anterior mid-gut anteriorly and laterally— the presumptive stomodaeum.

5. Postero-ventrally behind the posterior end of the presumptive mesoderm and in front of the pole cells—the presumptive posterior portion of the larval mid-gut.

6. At the posterior end, distinct from the blastoderm—the pole cells.

7. Surrounding the pole cells dorsally and laterally and abutting against the presumptive posterior mid-gut—the presumptive proctodaeum.

8. Occupying the lateral regions of the blastoderm and extending over the anterior end of the embryo—the presumptive ectoderm of the head and trunk.

9. Mid-dorsally as a strip stopping short of either end—the presumptive extra-embryonic ectoderm.

Ignoring for the moment the presumptive imaginal cells, which also find a place within the blastoderm, the events which result in these areas being separated out in the spatial arrangement from which organogeny proceeds will now be considered in turn.

As soon as blastoderm formation is complete the mid-ventral band of presumptive mesoderm begins to sink inwards, more rapidly in the midline than at the sides, while its lateral cells curve towards the midline and take up the space vacated at the surface by the median cells (Text-fig. 7 A, B; Plate 2, figs. A, B). Sinking in first takes place in the middle region of the embryo but rapidly spreads to the ends of the presumptive mesoderm and is obvious over the entire band by 8 hours (Text-fig. 6A). Within the next half hour the lateral borders of the gutter so formed approach each other in the midline so that it becomes an almost closed tube (Text-fig. 7, C–F). A connexion is retained, however, with the cells of the marginal strips. As the presumptive mesoderm sinks in, the marginal cells change from a columnar to a cuboidal form preparatory to contributing to the mesoderm itself.

Before mid-ventral closure is completed, i.e. between 8 and hours, celldivision begins within the mesoderm at a point about one-third of the way along its length (Text-fig. 7c) and, as closure of the groove proceeds, the divisions spread both forwards and backwards along the band. The peak of this mitotic activity is reached between and 9 hours. The number of cells in the band rapidly increases and the band itself elongates, growing round the posterior end of the yolk and forwards along its mid-dorsal surface to the boundary of the anterior third of the embryo (Text-fig. 6 B, c; Plate 2, fig. F). Elongation of the mesoderm in this way is part of a complex process of elongation involving almost all the presumptive areas of the embryo. Other aspects of the process are described below.

While the cells of the presumptive mesoderm are thus dividing, further cells are added to the band through cell division in the marginal strips (Text-fig. 8A). Finally, towards 9 hours, the edges of the mesodermal band come together, closing and obliterating the mid-ventral groove (Text-fig. 8 A, B) and leaving the band as an irregular multi-layered strip of polygonal cells, triangular in crosssection, extending from a point not far short of the anterior end of the embryo along the ventral midfine, round the posterior end and forward along the dorsal surface. At the anterior and posterior ends of the closing groove the marginal cells give rise to a surface epithelium which loses connexion with the mesoderm, but over most of its length closure of the groove results in formation by the marginal cells of a middle strand, some 3–4 cells wide, whose cells project inwards from narrow bases and enlarge at their inner ends so as to retain apparent continuity with the mesoderm.

After 9 hours, mitotic activity in the mesoderm decreases. Once the definitive band length is achieved at 11 hours, the mesoderm cells spread laterally beneath the embryonic ectoderm (see below) to form a single layer, completed by 12 hours after oviposition. Cell-division within the mesoderm is now infrequent. During the same period (9–12 hours), the mesoderm grows forward laterally on either side of the developing anterior mid-gut (see below) almost to the anterior end of the embryo.

The presumptive anterior mid-gut first becomes histologically distinguishable between and 9 hours when its cells, by amoeboid immigration, form the walls of a pear-shaped depression in front of the anterior end of the mid-ventral groove of the immigrating mesoderm. By 9 hours after oviposition cells have begun to be budded off from the apex and posterior surface of this depression (Text-figs. 6B; 8D; Plate 2, fig. C) and later, by 11 hours, also from its side walls, to form a mass of actively dividing small cells immediately internal to the depression itself. As the number of cells increases the mass grows posteriorly in the ventral midline so that by 12 hours it forms a wedge between the surface of the yolk and the mesodermal band (Text-figs. 6c; 8E; Plate 2, fig. F). At the same time, as described above, the anterior end of the mesodermal band grows forward on either side of the wedge. While these changes are taking place, closure of the mid-ventral groove immediately behind the anterior mid-gut rudiment transforms the horseshoe-shaped presumptive stomodaeum into a closed ring at the periphery of the anterior mid-gut depression (Text-fig. 8E).

The sequence of events through which the posterior mid-gut and proctodaeum complete their early development is more complex than the corresponding sequence for the anterior mid-gut and stomodaeum. The pole cells are closely associated with them, while elongation of the mesodermal band and its covering ectoderm (see below) also cause the movement of the posterior rudiments from their initial position to a final dorsal position at the border of the anterior third of the embryo. The movement begins about 8 hours after oviposition (Text-figs. 6A; 7B), proceeds rapidly as cell-division in the mesoderm reaches its peak between and 9 hours (Text-fig. 6B), continues at the same rate during the next hour, and then slows to completion by 11 hours. It can be followed in living de-chorionated eggs and a description of events at the egg surface forms a useful preliminary to a detailed description of development. As forward movement of the posterior complex begins, a shallow depression, circular in outline, with its posterior border drawn out to a point confluent with the mid-ventral mesodermal groove, forms around the pole cells. As the complex moves forward along the dorsal surface the depression deepens, swallowing the pole cells (Textfig. 1c), and by 10 hours, when the forward movement is almost complete, the rim is circular anteriorly and has a raised posterior transverse lip. During the last hour of its forward movement the aperture of the depression closes to a transverse slit with posteriorly curved lateral corners and a slightly raised anterior lip (Text-fig. ID).

The pole-cell group lies at first with its base in a cone-shaped cavity whose walls are blastoderm cells and whose floor is the surface of the yolk (Text-fig. 4c). The depression which forms around the pole cells results from elongation and inward migration of the blastoderm cells and initially no distinction can be made between any of these cells. Those immediately adjacent to the pole cells penetrate the most rapidly, so that the depth of the hollow cone in which the pole cells he increases and the pole cells sink inwards and are removed from the surface (Text-fig. 7B). By hours, however, the immigrating blastoderm cells immediately posterior and postero-lateral to the pole cells show a much greater penetration than their fellows (Text-fig. 7E). They are quickly displaced from the surface to the interior, the cells lateral to them closing over the surface to replace them. In this way the posterior mid-gut rudiment becomes histologically distinct as a mass of elongated cells posterior to the pole cells at the free inner end of the immigrating posterior complex. At the same time the horseshoeshaped presumptive proctodaeum is transformed into a ring encircling the opening of the depression in which the pole cells lie (Plate 2, fig. D). Completion of the ring is accompanied by closure of the posterior end of the mid-ventral mesodermal groove as described above.

The cells of the proctodaeal ring continue to migrate inwards during the next half hour, at the same time turning through 90° to become radially arranged as the wall of a proctodaeal tube, those which initially lay closest to the pole cells forming the inner end of the tube (Text-figs. 6B; 8C; Plate 2, figs. C, D). The anterior and lateral walls of the tube are established more quickly than the posterior, the formation of which is delayed until the posterior mid-gut cells have become internal.

The pole cells, sinking in more and more as the proctodaeum opens the way for them, remain a distinct mass occluding the end of the tube in contact with the yolk, with the posterior mid-gut rudiment lying behind them (Text-figs. 6B; 8c; Plate 2, fig. D). The proctodaeal tube now begins to increase in length, about 9 hours after oviposition, through division of the cells of its walls, while the posterior mid-gut rudiment begins to bud off a mass of small cells between the yolk and the posterior region of the mesodermal band. A few pole cells appear to disintegrate at this stage, since a small amount of granular material is always found in the proctodaeal lumen vacated by them, but the majority of the cells remain intact within the pole-cell mass. The condition described above persists until the end of gastrulation (Text-figs. 6c; 8E; Plate 2, fig. F), by which time the proctodaeum has reached its most anterior position. As a consequence of the simultaneous elongation and forward migration of the proctodaeum, its free inner end, together with the pole cells and posterior mid-gut rudiment, is displaced backwards through being dragged against the yolk.

By the end of gastrulation, the presumptive embryonic and extra-embryonic ectoderm have become converted into a surface layer enclosing the yolk and the various rudiments now internal.

The major part of the presumptive embryonic ectoderm forms two broad lateral bands of blastoderm cells and as mid-ventral immigration of the presumptive mesoderm proceeds these bands move ventrally as a whole (Text-figs. 7D; 8A). At about 8 hours, as the now internal mesodermal band begins to elongate, the posterior two-thirds of each band of ectoderm becomes narrower and presses down towards the ventral midline. As a result, the more posterior cells of the ectodermal bands move over the posterior pole of the embryo and forwards along its dorsal surface (Text-fig. 8 A, B; Plate 2, fig. C). Little mitotic activity is evident in the ectoderm during this time. The two bands, separated in the midline by the middle strand derived from the marginal strips (see above) thus join the elongating mesodermal band in pushing the posterior mid-gut-pole cell-proctodaeum complex into its antero-dorsal position. The anterior third of each ectodermal band is much wider than the posterior two-thirds, reaching almost to the dorsal surface, and does not appear to be involved in the process of elongation.

Elongation of the embryonic ectoderm is completed by 11 hours. Before this time, separation of rudiments within the bands begins, so that no clear distinction can be made between the end of gastrulation and the beginning of organogeny. In two ectodermal regions, on either side of the middle strand behind the anterior third of the embryo (Text-fig. 9E) and latero-dorsally on either side of the anterior third of the embryo, some of the original blastoderm cells show characteristic nuclear changes (enlargement of the nucleus, entry into prophase of mitosis) and migrate from the surface to the interior to lie between the surface cells and the cells of the mesodermal band. Here they begin to divide. These cells, as will subsequently be shown, are the neuroblasts from which the ganglia of the central nervous system arise. They form a row some 8 cells wide on either side of the ventral midline and a large patch on either side antero-dorsally. The cells which remain at the surface in their vicinity undergo some tangential divisions, maintaining a continuous surface layer without decrease in area.

The presumptive embryonic ectoderm which covers the anterior end of the embryo and borders the presumptive stomodaeum undergoes almost no change during gastrulation, remaining as a simple columnar epithelium (Text-fig. 8E). Towards 12 hours, however, scattered mitoses are seen in it, a preliminary to subsequent changes contributing to the formation of the head.

As the presumptive mesoderm begins to sink in mid-ventrally and the lateral embryonic ectoderm moves ventrally to replace it, the mid-dorsal blastoderm cells change from a columnar to a cuboidal form and the area of surface they occupy increases. The extra-embryonic ectoderm thus revealed has a clear boundary with the embryonic ectoderm and can be seen over the posterior two-thirds of the dorsal surface before forward migration of the posterior complex begins (Text-fig. 6A). While the embryonic ectoderm continues to move towards the ventral midline, the extra-embryonic ectoderm spreads laterally to replace it both through attenuation of its cells (Text-fig. 8A) and through celldivision. Then, as the posterior complex is pushed forwards along the dorsal surface, the extra-embryonic ectoderm is displaced from the dorsal midline (Text-fig. 6 B, c) and drawn back as two dorso-lateral horns which finally reach almost to the posterior end, separating the dorsal and ventral portions of the ectodermal bands (Text-fig. ID). Only at its anterior margin does the broad mid-dorsal area of extra-embryonic ectoderm persist in front of the proctodaeal anterior lip as a narrow transverse strip joining the two areas.

Over the anterior third of the embryo the dorsal blastoderm cells behave differently. The embryonic ectoderm of this region is broad, as described above, and shows little ventral displacement as the presumptive mesoderm sinks in. In association with this, no broad area of thinly spread epithelium is formed on the dorsal surface in this vicinity and only a narrow mid-dorsal strip of blastoderm becomes cuboidal and reveals itself as extra-embryonic ectoderm. The strip is continuous with the broader area posterior to it but develops later, at about 8 hours, and is not involved in the displacement of extra-embryonic ectoderm caused by forward migration of the posterior complex. It shows a small amount of cell-division during the early stages of its formation.

The close association between the activities of the embryonic and extraembryonic ectoderm during gastrulation is further reflected in the occurrence of cell-divisions in the embryonic ectoderm adjacent to the immigrating mesoderm and marginal strips over the anterior third and at the posterior end of each ectodermal band. It is precisely these regions which lack an attenuation and spread of the dorsal blastoderm as extra-embryonic ectoderm compensating for the loss of the mid-ventral cells from the surface. Such compensation is therefore gained through cell-division in the embryonic ectoderm adjacent to the mesoderm. The same process is seen at the lateral borders of the posterior mid-gutpole cell-proctodaeum complex when this begins to sink inwards (Text-fig. 7E). Anterior to the complex, compensatory surface coverage is gained from extraembryonic ectoderm formation; posteriorly, immigration of mesoderm is itself taking place.

It has been shown above that during gastrulation a distinction can be made between events taking place in the anterior third of the embryo and those in the posterior two-thirds. The former shows relatively little movement of cells over its surface while the latter shows the vigorous movements of ectoderm and mesoderm elongation and accompanying dorsal displacement of the posterior complex. Certain accessory processes, such as the relative growth of the extraembryonic ectoderm and the occurrence of cell-divisions in the ventral regions of the embryonic ectoderm during mesodermal immigration, confirm this distinction. It is therefore significant that at an early stage in gastrulation, as elongation of the embryonic ectoderm and mesoderm begins, a conspicuous fold appears in the surface of the embryo at the border between the regions whose subsequent behaviour differs. This fold, the cephalic furrow (Text-fig. 1c), is a deep intucking of the blastoderm into the yolk, first appearing laterally at 8 hours and rapidly extending dorsally and ventrally to encircle the embryo almost transversely, with a slight ventro-dorsal backward slope. That furrow formation is not a consequence of elongation is seen from the facts that the two events begin simultaneously and that the furrow is complete by hours, when elongation has hardly begun. The furrow persists through the early period of rapid elongation (Text-fig. 1c; Plate 2, figs. D, E), cell-division in the mesoderm beginning at the level of the furrow and spreading from here both forwards and backwards, but disappears between and 10 hours so that the final stages of elongation are completed without it (Text-fig. ID). Subsequent events reveal that the cephalic furrow lies approximately at the boundary between the future maxillary and labial segments of the head.

Behind the cephalic furrow a similar but less-conspicuous furrow develops, persisting through the same period of development (Text-fig. 1c). This second furrow does not cross the dorsal surface of the embryo. As elongation of the embryonic ectoderm and mesoderm proceeds, four further pairs of temporary lateral furrows form (Text-fig. 1c), but these are shallower and more shortlived than the first (cephalic) and second furrows, being evident only between and hours, i.e. during the first phase of rapid elongation.

As already described, the yolk at the end of blastoderm formation is enclosed by a cytoplasmic sac, anucleate except at the extreme posterior end, while a row of vitellophages occupies the longitudinal midline of the yolk mass (Text-fig. 3G). During gastrulation, the immigration of cells into the interior of the embryo displaces the yolk mass in a number of places (Text-fig. 6 A, B, C), but none of the cells enters the yolk itself. Formation of the cephalic furrow is also associated, perhaps causally, with a corresponding furrowing of the yolk. The vitellophages, on the other hand, move away from the midline, some becoming scattered through the yolk while the remainder migrate to the yolk surface and fuse with the perilecithal cytoplasmic sac, transforming it into a thin, nucleated epithelial yolk sac (Text-figs. 6 A, B, C; 8E).

No attempt is made below to present a full account of segmentation or to describe in detail the morphogenesis and histogenesis of the muscular system, tracheal system, peripheral nervous system, and sense organs. In general their development follows the course described by Poulson (1950) for Drosophila melanogaster. Particular attention is paid, however, to the development of the gut, Malpighian tubules and gonads, and to the fate of the extra-embryonic ectoderm.

As organogeny proceeds, the embryo undergoes a number of general changes typical of cyclorrhaphous Diptera.

Three hours after the end of gastrulation, the grooves which delineate the three post oral head segments, mandibular, maxillary, and labial, appear on the surface ventro-laterally on either side. After a further 3 hours, the limiting annuli of the trunk segments begin to form in antero-posterior succession as paired ventro-lateral transverse grooves, the 3 thoracic and 8 abdominal segments being delineated by the end of the 24 hours. At the same time, shortening of the germ-band occurs. The trunk segments concentrate on to the ventral surface, the proctodaeal opening is drawn back to its definitive position at the posterior end of the trunk, and the extra-embryonic ectoderm is redistributed over the dorsal surface. Within the next 4 hours dorsal closure is completed, the extra-embryonic ectoderm is resorbed, and the inter-segmental grooves extend dorsally and ventrally to encircle the body.

During shortening of the germ-band, involution of the head begins. The head lobe is pushed antero-dorsally by the crowding forward of the mouthpart segments and both head lobe and mouthpart segments are then infolded within the thoracic segments. Involution is completed in approximately 12 hours. Muscular activity soon follows, waves of contraction passing forwards along the trunk while the mouth hooks are repeatedly protruded and withdrawn. As development nears completion, activity increases, ultimately resulting in rupture of the vitelline membrane and chorion by the mouth hooks and escape of the larva through the tear.

Accompanying shortening of the germ-band, dorsal closure and involution of the head, the organ primordia separated out through gastrulation continue their development. The establishment of each major organ system from its primordia is described below.

Disregarding the complex cephalopharyngeal apparatus formed as a result of involution of the head, the gut of the newly hatched larva comprises a tube of varying diameter, with a lining epithelium covered externally by a thin coat of circular muscle. It can be divided into :

• The oesophagus, a thin tube with a cuticular lining, flattened epithelium and pronounced circular muscle coat, extending from the posterior end of the pharynx through the circum-oesophageal nerve ring.

• The proventriculus, into which the oesophagus continues as a muscular core, surrounded by a layer of pale-staining cells and an outer coat of cuboidal epithelium staining darkly with haematoxylin.

• The mid-gut, a convoluted tube whose epithelial wall is continuous anteriorly with the outer coat of the proventriculus. The mid-gut epithelium is cuboidal throughout.

• The rectum, beginning antero-laterally at the point at which the Malpighian tubules arise from the gut and looping forwards and upwards to run back in the dorsal midline above the coils of the mid-gut before turning down to the anus. The rectal epithelium is irregular and pale staining and lined by a thin cuticle.

The development of this elaborate tube from its four initial rudiments, the stomodaeum, proctodaeum, and anterior and posterior mid-gut, takes place as follows.

The stomodaeum (Text-figs. 9 A, E; 10) increases in length through celldivision in its epithelial wall, growing horizontally back beneath the yolk sac and then pushing up into it so that the surface of the yolk sac is infolded anteriorly. By the time shortening of the germ-band is complete, the stomodaeum has reached its definitive length, extending from the mouth to the first thoracic segment as a narrow tube with a cuboidal epithelial wall. As it grows the stomodaeum buds off two small dorsal diverticula, one behind the other. The fate of these has yet to be followed, but they resemble in origin the rudiments of the stomatogastric ganglia of Drosophila melanogaster (Poulson, 1950) and Calliphora erythrocephala (Ludwig, 1949).

The stomodaeum now differentiates directly into the oesophagus and the central core and intermediate cell layer of the proventriculus. With involution of the head, the embryonic mouth is carried inwards and the oesophagus displaced posteriorly to lie in the first and second abdominal segments.

As the stomodaeum begins to grow it pushes the enlarging wedge of anterior mid-gut cells before it (Text-fig. 9A), eventually as far as the first thoracic segment. At the same time, beginning at 14 hours, the wedge itself proliferates two lateral arms which embrace the yolk sac and grow latero-ventrally along it, reaching the posterior end of the embryo in approximately 3 hours (Textfigs. 9 c, D; 10). A second proliferation of cells occurs on the dorsal surface of the wedge between 14 and 15 hours, the products in this case penetrating into the yolk sac as tertiary vitellophages.

The proctodaeum (Text-figs. 9c; 10) also continues to grow through celldivision in its epithelial wall, pushing before it both the posterior mid-gut rudiment and the mass of pole cells. By 14 hours the pole cells can be separated into two distinct histological types. Sixteen of the cells show enlargement of the nucleus together with a cytoplasmic change which causes them to stain pinkishbrown with the stains employed, in contrast to the bluish-pink coloration of all other cells. The 16 cells then migrate out of the pole-cell mass to fie scattered beneath the enlarging posterior mid-gut rudiment (Text-fig. 9B). Their further fate as definitive germ-cells is described below. The remaining pole cells multiply and within a further 2 hours come to form the ventral wall of the distal end of the proctodaeum. Meanwhile the proctodaeum continues to grow and at 16 hours produces two pairs of diverticula, one pair laterally placed (Text-fig. 9c), the other dorsally placed immediately behind the first pair at the distal end of the tube. The further development of these, the rudiments of the Malpighian tubules, is described below.

Subsequent development of the proctodaeum is complicated by the onset of shortening of the germ-band during the 18th hour (Text-fig. 10). As the proctodaeal opening is drawn posteriorly, the elongating proctodaeum folds on itself to give a proximal arm running anteriorly beneath the extra-embryonic ectoderm and a distal arm running posteriorly beneath the proximal arm. This arrangement persists throughout shortening, the proximal arm growing more rapidly than the distal so that the inner end of the proctodaeum finally lies anterior to the anus. Once shortening is completed at 24 hours, the proximal arm continues its growth in the dorsal midline, pulling the distal arm forward to lie eventually in the 3rd and 4th abdominal segments. At the same time histodifferentiation of proctodaeal into rectal epithelium occurs and the epithelium finally secretes the lining cuticle of the rectum.

The posterior mid-gut rudiment (Text-fig. 9 B, D), occluding the distal end of the proctodaeum, is pushed back along the dorsal surface of the yolk sac, first as a result of proctodaeal growth, later as a combined result of proctodaeal growth and displacement due to shortening. At about 15 hours it begins to proliferate two lateral, posteriorly directed arms which grow dorso-laterally over the yolk sac and within 3 hours reach the posterior end and turn ventrally to meet the paired anterior mid-gut strands described above (Text-fig. 10).

The posterior ends of the paired mid-gut strands are carried posteriorly as a result of shortening so that by 24 hours the two strands stretch anteroposteriorly along the yolk sac. Within the next 4 hours, as dorsal closure is completed, the cells of the strands spread dorsally and ventrally over the yolk sac to enclose it in an ovoid epithelial sac stretching from the end of the stomodaeum to the end of the proctodaeum. Digestion of the yolk within the sac and of the yolk-sac wall and remaining vitellophages is accompanied by elongation and convolution of the sac itself and histodifferentiation of its wall as mid-gut epithelium. At the anterior end the cells of the mid-gut overgrow the end of the stomodaeum to form the outer epithelium of the proventriculus, while immediately behind the proventriculus the mid-gut epithelium is produced as four diverticula which give rise to the mid-gut caeca.

In the newly hatched larva two pairs of Malpighian tubules leave the gut at the junction of the mid-gut and rectum, one pair dorso-laterally and one ventrolaterally. The dorso-lateral pair coil posteriorly through the haemocoele, the ventro-lateral pair anteriorly. Each tube has a cuboidal epithelial wall and a small cylindrical lumen. Both pairs of tubules arise at 16 hours as diverticula of the distal end of the proctodaeum (see p. 269). As the gut develops, the bases of the tubules are carried with it to their definitive position in the 4th abdominal segment and at the same time each tubule grows through cell-division before undergoing histodifferentiation.

The salivary glands originate as two ventro-lateral plates of cells which first become distinct within the ectoderm of the labial segment at about 16 hours. Each plate invaginates and grows inwards as a blind-ending tube retaining a small surface aperture (Text-figs. 9E; 10). Growing first towards the dorsal surface, the inner end of each tube turns posteriorly to extend parallel to the stomodaeum through the thoracic segments. By 21 hours the vertical portion has differentiated as a thin duct with a flattened epithelium, the horizontal portion with its cuboidal epithelium remaining widely open. As involution of the head proceeds, the two ducts are drawn out and at the same time come together in the midline at the surface to gain a common opening. The inturning of the labial segment as part of the cephalopharyngeal apparatus brings this opening into the ventral wall of the latter, removing it from the body surface. The lumen of each gland becomes filled with secretion before hatching takes place. The salivary glands in Drosophila melanogaster develop in an identical manner (Sonnenblick, 1939, 1940, 1950; Poulson, 1950).

The development of the mesodermal bands after gastrulation has not been followed in detail, but a number of observations have been made. Significant among them are :

• the confirmation of separation of somatic and splanchnic mesoderm during the early stages of completion of the gut;

• the lack of any evidence of the development of segmentation in the mesoderm earlier than the final differentiation of segmental muscles of the body-wall. Segmentation of the embryo of D. tryoni appears to be confined to the embryonic ectoderm behind the mouth and all trace of somites within the developing mesoderm is lost.

The splanchnic mesoderm arises from cells of the anterior and posterior ends of the mesodermal bands which congregate around the stomodaeum and proctodaeum respectively during the first 6 hours of gastrulation (Text-fig. 10) and from two longitudinal strands of mesoderm which separate off from the paired mesodermal bands between 18 and 24 hours to lie alongside the distal end of the stomodaeum and the mid-gut strands (Text-fig. 9B). By the time shortening is complete the splanchnic mesoderm of the stomodaeum and proctodaeum has formed a thin epithelial layer over each, the future circular muscle layer. Then as the mid-gut strands spread over the yolk sac, the splanchnic mesoderm strands spread with them. From the resulting outer layer of cells arises the circular muscle of the mid-gut.

The somatic mesoderm shows little change between 12 and 24 hours other than cell multiplication and spread as a mesenchyme particularly in the head (Text-fig. 9, A–E). AS dorsal closure proceeds, however, it begins to histodifferentiate and by the time closure is complete it has given rise to the musculature of the segments and cephalopharyngeal apparatus and to the fat-body, conspicuous in the haemocoele on either side of the trunk as an irregular plate of vacuolated cells extending throughout the abdomen. A small number of somatic mesoderm cells also contribute to the developing gonads (see below). The larval heart arises from mesodermal cardioblasts.

The gonads of the newly hatched larva are a pair of ovoid bodies embedded in the fat-body dorso-laterally in the 5th abdominal segment. Sexes are indistinguishable, each gonad comprising 8 germ cells with large nuclei and prominent nucleoli, compacted together and covered by a thin epithelium. One or two small cells lie as interstitial cells among the germ cells.

As has already been pointed out, the 16 germ cells which enter the two gonads are modified pole cells which at 14 hours lie scattered beneath the developing posterior mid-gut rudiment (Text-fig. 9B). From this position 8 cells migrate on each side through the growing posterior mid-gut strands to enter the adjacent mesodermal band, so that by 18 hours they are lodged among the mesoderm cells of the 6th and 7th abdominal segments (Text-fig. 10). As shortening of the germ-band takes place, the germ cells are carried within the contracting mesoderm, at the same time moving farther anteriorly and towards each other to come together at 22 hours as a longitudinal row on each side adjacent to the now distinct splanchnic mesoderm in the 5th abdominal segment. Within the next 2 hours each row of germ cells compacts into an oval mass which becomes invested by a thin coat of somatic mesodermal cells. As dorsal closure and mesodermal differentiation proceed, the gonads are carried upwards to their definitive dorso-lateral positions adjacent to the developing fat-body. Beyond 28 hours they show no further change before hatching.

It has already been pointed out that the neuroblasts of the presumptive nervous system begin to separate from the embryonic ectoderm and to histodifferentiate and bud before the end of gastrulation. As this process continues, both the paired cerebral masses and paired longitudinal nerve-cords push deeply into the interior, displacing the mesoderm (Text-fig. 9, A–E). Superficial segmentation is not reflected in the developing central nervous system, neither part of which is segmentally subdivided. The paired ventral nerve-cords, however, follow the segmental annuli in developing in an antero-posterior sequence. By the time shortening of the germ-band begins the central nervous system is well developed and as shortening proceeds, it begins to separate from the surface layer of hypodermis and move into the interior. The middle strand, which remains distinct until this stage (Text-fig. 9E) is drawn in with it and subsequently incorporated into the ventral nerve-cord. Between 24 and 28 hours separation of the central nervous system from the hypodermis is completed and at the same time the two cerebral masses fuse in the midline above the oesophagus and on either side ventrally with the anterior end of the ventral nerve-cord. Segmentation also becomes apparent in the ventral nerve-cord in segmental fusion of the paired cords and the appearance within the ganglia of zones of neuropile finked by paired intersegmental commissures. Once isolated from the hypodermis, the central nervous system begins to condense, so that by hatching it is concentrated wholly within the first 4 abdominal segments and has lost its segmental pattern. Displacement of the circumoesophageal nerve-ring backwards to the second abdominal segment is a consequence of involution of the head. Development of the nervous system in D. tryoni is typical of Cyclorrhapha.

Typical of Cyclorrhapha, the larva of D. tryoni has two main longitudinal dorso-lateral tracheal trunks giving off small segmental branches and opening posteriorly by a pair of large postero-dorsal spiracles on the 8th abdominal segment. Later in larval life a second pair of large spiracles arises at the anterior ends of the tracheal trunks in the prothoracic segment. The origin of the tracheal trunks lies in paired lateral segmental invaginations of the embryonic ectoderm in the 3 thoracic and first 7 abdominal segments, arising simultaneously during the 16th hour (Text-fig. 9D). Each invagination grows anteriorly and posteriorly to link with those adjacent to it and at the same time breaks its connexion with the surface, estabfishing the paired trunks by the time shortening of the germ-band is complete. The posterior ends of the invaginations of the 7th abdominal segment grow posteriorly on either side of the proctodaeum to make contact with the developing hypodermis on either side of the anus, in which position the two definitive spiracles break through following dorsal closure.

The major part of the hypodermis develops directly out of the presumptive embryonic ectoderm, which undergoes little change other than scattered celldivisions until dorsal closure begins, but then shows cell-division and transformation from a columnar to a cuboidal epithelium as it spreads over the dorsal surface. Mid-ventrally and antero-dorsally on the head, hypodermal cells persist and spread at the surface as the central nervous system moves into the interior, so that by 28 hours a complete differentiated hypodermis covers the embryo. The segmental annuli, which first appear at 16 hours and are completed by 28 hours, originate as folds in the developing hypodermis. The special derivatives of the embryonic ectoderm of the head lobe and mouthpart segments which line the cephalopharyngeal apparatus have not yet been studied in detail.

Throughout its development, the extra-embryonic ectoderm shows no trace of extension as amnion or serosa. Its distribution at the end of gastrulation is held until shortening of the germ-band redistributes it as a band along the dorsal surface once more. With dorsal closure, it is resorbed, disappearing without trace by 28 hours.

The timetable of embryonic development in D. tryoni and the cell lineage of its larval organs (Text-fig. 11) are summarized below. The discussion which follows reviews aspects of dipteran embryology for which the present study provides new information.

Maturation of egg, fusion of pronuclei, formation of zygote nucleus.

1st and 2nd synchronous mitoses, 4 cleavage nuclei.

1–2 3rd-6th synchronous mitoses, 64 cleavage nuclei.

2–3 7th mitosis, invasion of blastoderm by 90 cleavage nuclei, 38 remaining as vitellophage nuclei; 8th, 9th, and 10th synchronous mitoses, 720 surface nuclei, 300 vitellophage nuclei.

3–4 11th and 12th synchronous mitoses of surface nuclei, giving 2,880; 4 pole cells cut off posteriorly, dividing once to give 8.

4–5 13th synchronous mitosis of surface nuclei, giving definitive number of 5,800. Pole cells divide once to give 16.

5–6 Syncytial blastoderm completed, vitellophages migrate to midline, pole cells divide once to give definitive number of 32.

6– Syncytial blastoderm becomes cellular, perilecithal cytoplasmic sac formed.

12 Gastrulation, separation of germ-band from extra-embryonic ectoderm, elongation of germ-band, invasion of perilecithal sac by vitellophages to form nucleated yolk sac.

12–17 Early stages of organogeny, delineation of mouthpart segments.

17–24 Shortening of germ-band, delineation of trunk segments, beginning of involution of the head.

24–28 Dorsal closure, resorption of extra-embryonic ectoderm, enclosure of yolk by mid-gut, continued involution of the head, rapid histodifferentiation.

28–32 Completion of involution of the head, continued histodifferentiation, elongation and convolution of mid-gut, beginning of condensation of the central nervous system.

32–42 Completion of histodifferentiation, condensation of central nervous system, resorption of yolk, onset of muscular activity, hatching.

In D. tryoni at 25° C., the first 10 synchronous cleavage mitoses are each completed in 15 minutes and the last 3 each in 30 minutes. This rate of cleavage is typical of known cases within the order. In Drosophila melanogaster each mitotic cycle in cleavage occupies 10 minutes at 25° C. (Huettner, 1933; Rabinowitz, 1941a; Sonnenblick, 1950) and in other Cyclorrhapha the blastoderm with its several thousand nuclei is completed within a few hours of oviposition (Reith, 1925; Auten, 1934; Scott, 1934; Fish, 1947; Hagan, 1951; Breuning, 1957; Imaizumi, 1958). Cleavage rates in the more primitive Nematocera have not been closely studied, but in Culex molestus, at 18° C., one mitotic cycle is completed every 20 minutes (Christophers, 1960) and in C. pipiens one every 15 minutes for the first 6 and one every 25 minutes for the remaining 5 cleavages (Idris, 1959, 1960).

In contrast to the relative constancy of cleavage rates in different species, the number of synchronous nuclear divisions taking place during cleavage and the timing of nuclear invasion of the egg surface are highly variable. In D. tryoni (p. 252), 7 synchronous divisions precede invasion of the surface and 6 follow, giving approximately 5,800 blastoderm nuclei. In Drosophila melanogaster, 8 synchronous divisions precede invasion of the surface (Huettner, 1935) and 4 follow (Sonnenblick, 1950), giving 3,200 blastoderm nuclei. No satisfactory data are available for other Cyclorrhapha, but similar variations are seen among the Nematocera. Christophers (1960) has recently shown that in C. molestus nuclear invasion of the surface follows the 10th synchronous cleavage mitosis, when approximately 720 nuclei move outwards and undergo 3 further synchronous cleavages to give between 5,000 and 6,000 blastoderm nuclei. In C. pipiens, in contrast, invasion of the surface begins after the 7th cleavage and the 120 nuclei which move outwards undergo 4 further cleavages to give about 2,000 blastoderm nuclei (Idris, 1959, 1960). In Sciara coprophila (Du Bois, 1932) and Wachtliella persicariae (Geyer-Duszinska, 1959) nuclear invasion of the surface again follows the 7th and in Miastor metraloas (Kahle, 1908) and M. americana (Hegner, 1912) the 6th cleavage, but the final number of blastoderm nuclei is unknown.

The differences in cleavage discussed above are not accompanied by differences in the mode of blastoderm formation. Indeed, blastoderm formation in D. tryoni (p. 255) is identical with that in Drosophila melanogaster (Child & Howland, 1933; Poulson, 1937; Rabinowitz, 1941tz; Sonnenblick, 1947, 1950) even in such details as the simultaneous appearance of nucleoli in the blastoderm before cellularization begins and the cutting off by the inner tangential boundaries of the blastoderm cells of a cytoplasmic sac around the yolk. In other Cyclorrhapha, blastoderm formation is essentially similar, although the cytoplasmic sac of Dacus and Drosophila has not been described for other species. In Musca domestica (Reith, 1925), Calliphora vomitoria (Kowalevsky, 1886; Voeltzkow, 1888, 1889), C. erythrocephala (Graber, 1889; Noack, 1901; Pauli, 1927), Lucilia sericata (Fish, 1947), and Phormia regina (Auten, 1934) the syncytial blastoderm shows distinct outer and inner layers, the secondarily incorporated material of the cytoplasmic reticulum being separated from the initial material of the cytoplasmic haloes by a thin layer of yolk, but the latter is absorbed during gastrulation and appears to be of little importance. In Nematocera, little attention has been paid to the details of blastoderm formation in any species, but the work of Idris (1960) indicates a close similarity in this respect between Culex pipiens and the Cyclorrhapha.

The cutting off in D. tryoni of pole cells which incorporate the posterior pole plasm and polar granules (p. 254) and lose mitotic synchrony with the cleavage nuclei is a characteristic dipteran developmental feature, but the number and mode of origin of the pole cells varies between species. D. tryoni differs from other Cyclorrhapha in always producing a fixed number of pole cells, 4, which are cut off during the 12th cleavage and subsequently divide in synchrony three times to give 32. In Drosophila melanogaster (Huettner, 1923; Geigy, 1931; Poulson, 1937; Rabinowitz, 1941tz; Sonnenblick, 1950) a variable number of nuclei, between 3 and 11, enter the posterior pole plasm and are cut off during the 8th-9th cleavages in pole cells which multiplyin an irregular manner to give between 36 to 73 cells (Rabinowitz, 1941zz). Calliphoraerythrocephala (Noack, 1901), C. vomitoria (Kowalevsky, 1886), Luciliasericata(Fish, 1947), stnáPhormiaregina (Auten, 1934) also initially cut off a variable number of cells and the only known parallel to the condition in D. tryoni is in Melophagus ovinus (Lassman, 1936), which initially cuts off 12 pole cells. In the Nematocera, pole cell origin is always precise but there is no constancy in its timing or in the number of cells in different species. In Culex pipiens (Idris, 1959, 1960) 6 cells are cut off following the 7th cleavage and divide once in synchrony to give a group of 12. In Sciara coprophila (Du Bois, 1932), 2 cells are cut off after the 6th cleavage and divide several times to give a group of 22–28. Two cells are also cut off in Phytophaga destructor, but after the 4th cleavage, then dividing to give a group of 16 (Metcalfe, 1935). In Chironomus a single cell is cut off after the 2nd (Weismann, 1863; Ritter, 1890; Hasper, 1910,1911; Yajima, 1960), and in Miastor after the 3rd cleavage (Kahle, 1908; Hegner, 1912, 1914; Nicklas, 1959) and divides three times to give a group of 8.

While the blastoderm and pole cells are forming, a number of cleavage nuclei and associate cytoplasmic haloes are always set aside as vitellophages within the yolk. Their mode of origin and final number in D. tryoni, however, is far from characteristic. In this species (p. 253), about 38 of the 128 nuclei resulting from the 7th cleavage remain behind in the yolk as the others invade the egg surface and undergo 3 (sometimes 4) further mitoses in synchrony with the surface nuclei to give approximately 300 (sometimes 600) primary vitellophages. At the same time, 2-5 of the nuclei which invade the posterior pole of the eggs are left behind in the cytoplasmic sac at the yolk surface when the pole cells are cut off, and of these, one or two reinvade the yolk with cytoplasmic haloes as secondary vitellophages. In Drosophila melanogaster, in contrast, nuclear counts made by Sonnenblick (1950) indicate that approximately 50 of 256 nuclei produced at 8th cleavage remain with their cytoplasmic haloes in the yolk. Since Poulson (1950) puts the definitive number of primary vitellophages at about 100, it appears that the nuclei undergo only one synchronous mitosis after separation, though Poulson agrees with Rabinowitz (1937, 1941a) that they then become polyploid through a single endomitosis. Rabinowitz (1941b) also records reinvasion of the yolk by a number of blastoderm cells as secondary vitellophages, but Poulson was unable to confirm this. Between 20 and 30 pole cells, however, wander back between the blastoderm cells into the yolk (Rabinowitz, 1941 a, b;Sonnenblick, 1941, 1950; Poulson, 1947, 1950) and are generally interpreted as secondary vitellophages (tertiary vitellophages of Rabinowitz), though Poulson (1950, 1959) and Poulson & Waterhouse (1960) have suggested an alternative fate for them (compare p. 284).

In other Cyclorrhapha, primary vitellophages similar in origin to those of Dacus and Drosophila occur in Glossina tachinoides (Hagan, 1951), Lucilia sericata (Fish, 1947), Phormia regina (Auten, 1934), and Melophagus ovinus (Lassman, 1936), while in Phormia and Melophagus in addition and in Musca domestica (Reith, 1925), Calliphora vomitoria (Voeltzkow, 1888, 1889), and C. erythrocephala (Noack, 1901) exclusively, they arise by reinvasion of the yolk by blastoderm cells (sometimes called secondary vitellophages). Secondary vitellophages like those of Dacus and Drosophila have also been reported in Calliphora, Lucilia, and Melophagus. In Nematocera, little attention has been paid to vitellophage origin. Christophers (1960) describes approximately 550 primary vitellophages remaining in the yolk in Culex molestus, but in C. pipiens (Idris, 1959, 1960) only 32 vitellophages are so formed. In Sciara coprophila (Du Bois, 1932) and Miastor metraloas (Kahle, 1908) a small number of vitellophages is formed by reinvasion of the yolk by blastoderm cells, as in Musca, &c.

Thus while the rate of synchronous cleavage and the mode of formation of the blastoderm vary little among species, no fixed pattern can be discerned in the number of cleavage divisions, the timing of nuclear invasion of the egg surface and the number and mode of origin of the pole cells and vitellophages.

It has already been shown (p. 256) that the complexities of gastrulation in Diptera are best approached by preliminary reference to the presumptive areas of the cellular blastoderm. The detailed resemblance of the presumptive area map in D. tryoni to that presented by Poulson (1950) and recently confirmed experimentally in part by Hathaway & Selman (1961) for Drosophila melanogaster is suggestive of a constant pattern of blastodermal presumptive areas in Diptera and a concomitant uniformity of gastrulation processes throughout the order is discernible from a comparison of gastrulation in D. tryoni and other species. As the following discussion shows, only minor differences are found and many of these appear to be differences in interpretation of essentially similar processes.

The concentration in D. tryoni of the embryonic presumptive rudiments within the ventral and lateral areas of the blastoderm as a germ-band and the elongation of the germ-band during gastrulation in such a way that its posterior end pushes forward along the dorsal surface of the embryo has been noted in all described cases. A much less extensive simultaneous extension of the anterior end of the germ-band over the anterior pole occurs in Drosophila melanogaster (Ede & Counce, 1956), Glossinia tachinoides (Hagan, 1951), Melophagus ovinus (Hardenberg, 1929), Sciara coprophila (Butt, 1934), and Simulium pinctipes (Gambrell, 1933) but not in D. tryoni and other species.

The mid-ventral presumptive mesoderm in D. tryoni behaves in essentially the same manner during gastrulation as in Drosophila melanogaster (Sonnenblick, 1950; Poulson, 1950) and other species, invaginating and subsequently spreading as paired ventro-lateral mesodermal bands. In both Dacus and Drosophila elongation of the invaginated band of mesoderm takes place through rapid cell-division following occlusion of the invagination, but the initiation of mitotic activity at the level of the cephalic furrow (see below, p. 282) in the former species is not recorded in the latter. The marginal strips which in D. tryoni (p. 260) contribute to the mesoderm before closure of the mid-ventral invagination are also unknown for other species.

The anterior mid-gut rudiment in D. tryoni (p. 260) is distinct from the mesoderm both histologically and in time, since it begins to invaginate only when mesodermal invagination is almost complete and then sinks inwards and buds off a mass of small dividing cells which pushes back between the mesoderm and the yolk. A similar development of the anterior mid-gut rudiment is described by Poulson (1950) for Drosophila melanogaster. Other accounts generally fail to distinguish early in gastrulation between the mesoderm and the anterior and posterior mid-gut rudiments, but the pos.tion of the anterior mid-gut rudiment and the general indications given of its or.’gin suggest a close similarity to Dacus and Drosophila in every case.

The early development of the stomodaeum has been little studied in Diptera. Its initial position in D. tryoni as a horseshoe-shaped presumptive area enclosing the presumptive anterior mid-gut anteriorly and laterally and its subsequent closure as a ring at the periphery of the anterior mid-gut invagination are repeated in Drosophila melanogaster (Poulson, 1950), save that in this species the stomodaeal rudiment closes across the mouth of the anterior mid-gut invagination as a stomodaeal plate. In other species the rudiment appears to occupy the same site, although its position relative to the anterior end of the embryo depends on the extent to which it is carried forwards during elongation of the germ-band. In Melophagus ovinus (Lassman, 1936), for example, it lies antero-dorsally at the end of gastrulation.

The development of the posterior mid-gut rudiment must be considered together with that of the proctodaeal rudiment and pole cells with which it is associated. Events in D. tryoni (p. 260) are particularly clear cut, since the posterior mid-gut can be distinguished from the proctodaeum as soon as the two begin to move into the interior and the establishment of the posterior midgut rudiment and pole cells at the inner end of the elongating proctodaeal tube is easily followed. The pole cells, due to their initial position, retain contact with the yolk surface throughout. In Drosophila melanogaster (Sonnenblick, 1950; Poulson, 1950) there are several points of difference. The presumptive posterior mid-gut in this species occupies the entire area of blastoderm beneath the pole cells. During gastrulation it forms a deep invagination into which the pole cells remaining after separation of the secondary vitellophages (p. 278) sink, separated from the yolk by the posterior mid-gut cells. The proctodaeum is then formed from cells at the rim of the invagination as a short tube which pushes the invagination more deeply into the interior. In earlier accounts (Poulson, 1937; Rabinowitz, 1941a; Sonnenblick, 1941) the initial invagination was also interpreted as proctodaeal, with posterior mid-gut cells only at its tip, since the Malpighian tubules which in other Diptera have always been assigned a proctodaeal origin (as in D. tryoni, see p. 269) arise in Drosophila from the walls of the initial invagination. Its interpretation as posterior mid-gut is probably to be accepted on the grounds that adequate experimental analysis would reveal presumptive Malpighian tubule cells in the blastoderm in slightly different positions relative to the presumptive posterior mid-gut and proctodaeal cells in Drosophila and Dacus.

While available accounts of the development of the posterior complex during gastrulation in other Diptera, especially that for Melophagus ovinus by Lassman (1936), suggest that events in D. tryoni and Drosophila melanogaster are typical of Cyclorrhapha, it is difficult in most cases to follow the development of each separate rudiment. Re-examination by modern methods, however, would probably reveal in every species a posterior mid-gut rudiment distinguishable early in gastrulation and a proctodaeum invaginating as a tube at the rim of the posterior mid-gut and pushing it more deeply into the interior.

In Nematocera, the posterior mid-gut rudiment as an invagination at the posterior end of the germ-band and the proctodaeum as a secondary invagination at the same place have been briefly described in Culex molestus (Christophers, 1960), C. pipiens(Idris, 1959, 1960), Aedes dorsalis (Telford, 1957), Sciaracoprophila (Du Bois, 1932; Butt, 1934), Miastor metraloas (Kahle, 1908), and Simuliumpinctipes (Gambrell, 1933), but a detailed study is wanting. More is known, however, of the behaviour of nematoceran pole cells. Only in C. molestus (Christophers, 1960), C. pipiens (Idris, 1959, 1960), and A. aegypti (Gander, 1951) are they known to be carried along the dorsal surface of the embryo at the end of the germ-band before sinking into the interior. In other species they migrate inwards before gastrulation begins, either directly into the yolk as in Sciara coprophila (Du Bois, 1932), or between the cells of the blastoderm into the yolk (Chironomus confinis—Hasper, 1910, 1911; Miastor—Kahle, 1908; Hegner, 1912; Phytophaga destructor—Metcalfe, 1935; Wachtliella persicariae—GeyerDuszinska, 1959; Simulium pinctipes—Gambrell, 1933; and probably Culex tarsalis—Rosay, 1959), move through the yolk and separate as two groups which come to rest antero-dorsally on either side of the ingrowing posterior midgut and proctodaeum. Their further fate is discussed below (p. 285). It is interesting that the initial polar group of nuclei of the cytoplasmic sac around the yolk in D. tryoni (p. 256) and some of the secondary vitellophages arising from the pole cells in Drosophila melanogaster (p. 278) also move forwards during germ-band elongation and come to rest below the posterior mid-gut rudiment. A change of fate is here evidently associated with retention of primitive behaviour.

In every described species of nematoceran and cyclorrhaphan, as in D. tryoni (p. 263) the embryonic ectoderm, which lies lateral and anterior to the invaginating mid-ventral rudiments, remains as columnar blastoderm throughout gastrulation (save for the histodifferentiation of neuroblasts towards the end of this process, p. 263), while the extra-embryonic ectoderm differentiates by attenuation and spread of its cells as a thin dorsal epithelium meeting the embryonic ectoderm anteriorly and laterally and the anterior proctodaeal wall posteriorly. As the germ-band elongates, the extra-embryonic ectoderm is displaced in part from the dorsal surface and drawn back as two dorso-lateral horns separating the ventral (anterior) and dorsal (posterior) portions of the germ-band. Closure of the edges of the mesodermal invagination brings the two lateral bands of embryonic ectoderm together mid-ventrally and where they meet, a histologically distinct middle strand is established over almost the entire distance between the stomodaeal and proctodaeal rudiments (D. tryoni, see p. 260; Drosophila melanogaster—Poulson, 1950; Lucilia ceasar—Escherich, 1902 a, b). The cells which form the middle strand in D. tryoni are those marginal strip cells which have already contributed to the mesoderm before mid-ventral closure (pp. 257 and 260). The middle strand is subsequently incorporated into the ventral nerve-cord.

Elongation of the germ-band involves an increase in length of the lateral bands of embryonic ectoderm. The cell activities which bring this about and at the same time cause, in combination with elongation of the mesoderm, forward displacement of the posterior mid-gut-proctodaeal-pole cell complex, have been considered only in D. tryoni (p. 263), Drosophila melanogaster (Sonnenblick, 1950; Poulson, 1950) and Calliphora erythrocephala (Breuning, 1957). In D. tryoni it is plain that crowding of the presumptive ectoderm cells towards the ventral midline behind the cephalic furrow is largely responsible for elongation of the ectoderm, displacement of the posterior rudiments, and redistribution of the extra-embryonic ectoderm. Throughout gastrulation, cell-division in the embryonic ectoderm is rare. Only where, as described on p. 264, mid-ventral crowding is at a minimum, does cell-division play a part in ectodermal spread, i.e. in surface replacement of the immigrating posterior complex and at the edges of the mid-ventral groove in front of the cephalic furrow. A number of statements by Sonnenblick and by Poulson indicate that the same processes take place in Drosophila. In particular, cell-division again appears to play a minor role in ectodermal redistribution, being confined, as in D. tryoni, mainly to the area around the immigrating posterior complex. Breuning (1957) has attributed an exaggerated importance to this activity, implying that it is a prime factor, together with mesodermal cell-division, in elongation of the posterior part of the germ-band. This view is necessitated by her interpretation from surface views of elongation of the germ-band in C. erythrocephala through activity of a posterior growth zone. Since the posterior region of ectodermal mitotic activity in Drosophila and Dacus is not a growth zone in Breuning’s sense, this interpretation must be regarded with caution until more detailed histological studies have been made. It seems more probable that the ectodermal gastrulation movements of Dacus and Drosophila will be found to be typical of all species and that, as Idris (1959) has recently shown for Culex pipiens, elongation of the germ-band takes place over its entire length behind the cephalic furrow.

The temporary furrowing of the surface of the embryo which accompanies gastrulation and simultaneous elongation of the germ-band in D. tryoni (p. 265) is characteristic of dipteran embryos, but widely different patterns of furrowing occur in different species and the interpretation of their significance by individual authors is by no means constant. The cephalic furrow appears to be characteristic of Cyclorrhapha and is also present in Culex pipiens (Idris, 1959, 1960) and C. tarsalis (Rosay, 1959). For Drosophila, Poulson regards it as lying at the boundary between head and trunk and functioning as a first step in differentiation between the two, but in D. tryoni (p. 266), Calliphora erythrocephala (Breuning, 1957) and Culex pipiens (Idris, 1959, 1960) it lies between the future maxillary and labial segments. This more or less arbitrary position, together with its initial appearance at the onset of germ-band elongation, its effacement when the latter is almost complete and the fact that the majority of germ-band elongation takes place posterior to it, lends support to the view put forward by Sonnenblick, that it acts as a stabilizing factor against which the forces involved in elongation of the germ-band operate. How the furrow itself forms is not clear.

It is notable that all other temporary furrows form behind the cephalic furrow during germ-band elongation and that their arrangement varies widely among species. In D. tryoni, the second transverse furrow behind the cephalic furrow may assist the action of the former, but the four pairs of more transient ventrolateral furrows seem to be a simple consequence of buckling during elongation. Similar folds appear in C. pipiens (Idris, 1959, 1960) and C. tar salis (Rosay, 1959). In Drosophila melanogaster such buckling is confined to the extraembryonic ectoderm in front of the forward-pushing posterior end of the germband, several initial folds forming which finally amalgamate as a single deep mid-dorsal fold, soon disappearing. In Calliphora erythrocephala (Breuning, 1957) an extreme regularity characterizes the additional temporary folds, which appear in rapid antero-posterior succession in the extra-embryonic ectoderm behind the cephalic furrow as five transverse folds persisting until elongation of the germ-band is almost complete. The earlier workers on Calliphora and Lucilia, who described similar additional folds, considered them to be a mechanical consequence of germ-band elongation, as in Drosophila and Dacus, but Breuning, on the basis of their regularity and the fact that the cephalic fold is inter-segmental in position, postulates that they represent the inter-segmental boundaries of the first 5 trunk segments. It has already been shown that Breuning’s assumptions about the mode of growth of the trunk in Calliphora, which are prerequisite to this interpretation of the temporary folds, are unwarranted (p. 282), so that her view of the folds itself must probably be rejected. More needs to be known of the stresses set up by the forces involved in gastrulation in Diptera before a satisfactory explanation of surface furrowing can be expected.

The changes which take place in the yolk in D. tryoni during gastrulation, mainly passive changes of shape, are similar in all species. A causal connexion has been postulated between yolk contraction and cephalic furrow formation in Drosophila melanogaster (Sonnenblick, 1950) but there is no experimental evidence on this point. The migration of many of the vitellophages to the yolk surface during gastrulation to fuse with the cytoplasmic sac and transform it into a nucleated yolk sac in D. tryoni (p. 266) is also known for Drosophila melanogaster (Poulson, 1950) and occurs after gastrulation in Miastor metraloas (Kahle, 1908).

The organogeny of the larval organs in Diptera has been subjected to detailed study only by Poulson (1950) for Drosophila melanogaster and only aspects of it for which D. tryoni has provided additional data will be reviewed here. The general changes, such as external delineation of segments, shortening of the germ-band, dorsal closure and involution of the head, which accompany organogeny in D. tryoni in a manner typical of cyclorrhaphous Diptera, have already been discussed (p. 266).

The development of the gut in D. tryoni (p. 268) from four initial rudiments, stomodaeum, proctodaeum, and anterior and posterior mid-gut, is typical of Diptera. The only point of controversy is the contribution made by the pole cells in Cyclorrhapha to the wall of the mid-gut. As already stated, they always lie at the end of gastrulation in the wall of (D. tryoni) or the lumen of (Drosophila melanogaster and other species) the posterior mid-gut-proctodaeal complex. From here, some of the pole cells migrate outwards as definitive germ-cells (see below, p. 286); some remain associated with the posterior mid-gut rudiment. It was generally assumed by earlier workers that these degenerated, but Poulson (1947, 1950) showed and with Waterhouse (1960), confirmed experimentally for Drosophila melanogaster that they are carried inwards by the tips of the growing posterior mid-gut strands and give rise to the middle region of the mid-gut. Poulson & Waterhouse also demonstrated a pole-cell origin of midgut cells in Lucilia cuprina and expressed the opinion that in Drosophila all the pole cells entering the embryo via the posterior mid-gut invagination have such a fate, though it now seems (see below, p. 286) that this is an exaggerated view. In D. tryoni (p. 269) the pole cells remaining in contact with the posterior midgut rudiment also contribute to the gut-wall, but without change of position, forming part of the wall of the proctodaeum. Reference to the presumptive fate maps of Drosophila and Dacus explains both the difference in pole-cell behaviour in the two species and the fact that the pole cells make any contribution to the gut-wall. In the Dacus blastoderm the cells adjacent to the pole cells are mainly proctodaeal, in Drosophila exclusively posterior mid-gut. Since determination in cyclorrhaphan eggs takes place very early, possibly before fertilization (Reith, 1925; Pauli, 1927; Anderson, 1961; Hathaway & Selman, 1961) and the fate of the presumptive areas of the blastoderm is fixed presumably by biochemical differentiation within the periplasm of the egg, it is easy to see that the boundaries of the periplasmic areas might overlap morphologically distinct parts of the blastoderm into which they are subsequently incorporated.

The vitellophages and the yolk sac to which they contribute in Dacus, Drosophila, and Miastor appear in general to be totally digested with yolk once the mid-gut wall is complete, although Poulson (1950) suggests that some of the cells of the yolk-sac wall in Drosophila melanogaster become incorporated into the mid-gut wall. There is no evidence of this in D. tryoni (p. 270).

The contribution of the paired mesodermal bands to the morphology of the larva in D. tryoni (p. 271) is typical of Diptera, conforming essentially to the pattern described in Drosophila melanogaster by Poulson (1950), but the question of the segmental composition of the dipteran larva required attention, since it has recently been raised by Butt (1960). It is convenient for purposes of discussion to consider the trunk separately from the head.

In the developing trunk, segmentation is established first in the ectoderm in antero-posterior succession (e.g. D. tryoni, pp. 266 and 273). Sciara coprophila (Du Bois, 1932; Butt, 1934) and Miastor metraloas (Kahle, 1908) show subsequent segmentation of the mesodermal bands into paired somites. In Simulium pinctipes (Gambrell, 1933), however, paired strands of splanchnic mesoderm separate from the mesodermal bands before the somatic mesoderm segments. Drosophila melanogaster (Poulson, 1950) resembles Simulium in that only the somatic mesoderm segments, while in D. tryoni (p. 271) even the basic segmentation of the somatic mesoderm is lost and all segmental derivatives of the mesoderm form by direct association with the segmented ectoderm during histodifferentiation. There is thus a tendency for the basic mesodermal segmentation in the trunk of Cyclorrhapha to be lost.

In the head, external segmentation can be recognized with certainty only for the mouth part segments, which in D. tryoni, as in all species, become externally delineated before segmentation of the trunk begins. No corresponding segmentation of the mesodermal bands occurs, while in front of the mouth neither ectoderm nor mesoderm retains vestiges of primary segmentation (paired somites and associate segmental ganglia—Manton, 1949; Anderson 1959) in any species whose embryology is known (this account p. 271; Ludwig, 1949; Poulson, 1950; Breuning, 1959; &c.). The conclusion of Butt (1960) that the anterior head lobe described by Breuning (1957) for the embryo of Calliphora erythrocephala represents the labrum of the dipteran larval head is therefore unfounded, since the pre-oral composition of the head cannot be segmentally analysed.

It has been shown for several species that the germ cells of the rudimentary larval gonads take origin from pole cells while the gonad sheath and interstitial cells take origin from mesoderm. D. tryoni conforms to this pattern (p. 272) and also resembles other Cyclorrhapha in that certain of its pole cells become incorporated in the wall of the gut (p. 269). In Nematocera, all the pole cells become germ cells (Ritter, 1890; Kahle, 1908; Hasper, 1910, 1911; Hegner, 1912, 1914; Du Bois, 1932; Gambrell, 1933; Metcalfe, 1935; Idris, 1959, 1960). The explanation of the dual fate of the cyclorrhaphan pole cells in terms of the relative timing of the determination of presumptive areas in the periplasm and formation of the pole cells and blastoderm cells can be extended to explain the difference in fate between nematoceran and cyclorrhaphan pole cells if account is taken of recent work by Yajima (1960) on determination in Chironomus dorsalis. In this species determination is not completed until the syncytial blastoderm stage, whereas the single initial pole cell is cut off after the second synchronous cleavage (cf. p. 277). There is thus no possibility that the presumptive areas of the blastoderm might overlap the pole-cell-forming region. The regular early formation of pole cells in Nematocera (p. 277) and their strict subsequent development as germ cells indicates similarly late determination for other members of the group, although the work of Idris (1959) on Culex pipiens suggests that the Culicidae may be closer to the Cyclorrhapha in this respect.

A pole-cell origin of germ cells has been described in Cyclorrhapha for Drosophila melanogaster, Calliphora erythrocephala, Lucilia cuprina, and Melophagus ovinus as well as for the present case of D. tryoni. The migration in the latter of 16 pole cells from the posterior mid-gut-pole cell-proctodaeal invagination to the definitive gonad positions (p. 272) reflects the condition in C. erythrocephala (Noack, 1901) and M. ovinus (Lassman, 1936), although in these two species it is not yet clear how many cells move to the gonads and how many remain behind, nor what is the fate of the latter. For L. cuprina, evidence of the polecell origin of germ cells at present rests with the experimental results of Poulson & Waterhouse (1960) and the actual migration of the cells has not been described. The same series of experiments also forms the basis of assertions by these two authors that in Drosophila melanogaster the germ cells arise from pole cells re-entering the yolk at the posterior pole before gastrulation begins and that the pole cells entering via the posterior mid-gut invagination during gastrulation form mid-gut cells only. This contradiction of the classical account of germcell origin in D. melanogaster, which describes pole cells migrating to the definitive gonad positions after entry via the posterior mid-gut invagination (Huettner, 1940; Sonnenblick, 1941,1950; Aboim, 1945) in the same manner as in D. tryoni, has recently been refuted by Hathaway & Selman (1961) and it can now be accepted that D. tryoni and Drosophila melanogaster are closely similar and characteristic of Cyclorrhapha in their mode of germ-cell formation.

The extra-embryonic ectoderm which differentiates by flattening and attenuation of the dorsal cells of the blastoderm and is subsequently displaced in large part from the dorsal surface by the elongating germ-band, undergoes no further changes in D. tryoni other than return to the dorsal surface when the germ-band shortens and subsequent resorption when dorsal closure takes place. Such extreme reduction of the extra-embryonic tissue is typical of Cyclorrhapha, where the only suggestion of amnion formation is in the temporary indrawing of the edge of the extra-embryonic ectoderm by the invaginating proctodaeum in Drosophila melanogaster (Poulson, 1950). In Nematocera, in contrast, the extra-embryonic ectoderm extends over the germ-band during gastrulation as a typical amnion and serosa. The reasons for the extreme reduction of the embryonic membranes in Cyclorrhapha are obscure and will no doubt remain so until the functions of these membranes in Diptera have been elucidated.

The general features of cell lineage of the larval organs in Diptera are more or less constant, as a comparison of the cell-lineage plan presented by Poulson (1950) for Drosophila melanogaster with that given for D. tryoni on p. 274 shows. The major difference between the cell-lineage plan given here for Dacus and that of Poulson, however, is the introduction of the concept of presumptive areas. The tentative adoption of this interpretation is a necessary step in formulating hypotheses on which experimentation can be based. If the theoretical concept of three germ-layers—ectoderm, mesoderm, and endoderm—is retained, the design of experiments is limited by the question, what are the differences between the primordia of the three layers? Greater progress will probably be made if the fundamental questions become (a) What are the differences between the presumptive areas of the blastoderm? (b) How are these differences established? (c) What do they lead to in terms of future interaction or autonomous development?

1. L’œuf de D. tryoni subit 7 divisions nucléaires synchrones à l’issue desquelles 90 noyaux colonisent la surface. Ces noyaux de surface subissent alors 6 autres divisions synchrones, réalisant ainsi un blastoderme de 5.800 cellules. Les 38 noyaux demeurés dans le vitellus se divisent 3 fois de façon synchrone, donnant environ 300 vitellophages primaires. Au pôle postérieur 4 cellules polaires se séparent, qui se divisent 3 fois de façon synchrone, en donnant 32 cellules. De 2 à 5 noyaux forment au pôle postérieur les vitellophages secondaires.

2. La gastrulation, impliquant l’invagination des ébauches antérieure et postérieure de l’intestin moyen, s’accompagne de l’invagination du mésoderme médioventralement, et des cellules polaires ainsi que du proctodeum. Le blastoderme dorsal s’amincit en ectoderme extra-embryonnaire reporté latéralement lors de l’allongement de la bandelette germinative qui a lieu au cours de la gastrulation. L’allongement se produit principalement en arrière d’un sillon céphalique temporaire. Plus en arrière apparaît un autre sillon transverse, puis 4 paires de replis latéraux également temporaires. La plupart des vitellophages migrent à la surface du vitellus pour former un sac vitellin nucléé.

3. L’organogenèse s’accompagne de l’apparition des segments, du raccourcissement de la bandelette germinative, de la fermeture de la région dorsale et de l’involution de la tête. Le tube digestif se développe à partir du stomodeum, du proctodeum, et des ébauches de l’intestin antérieur et de l’intestin postérieur. Les tubules de Malpighi apparaissent comme des expansions du proctodeum et les glandes salivaires comme des plaques ectodermiques ventro-latérales du segment labial. Le mésoderme splanchnique donne la musculature viscérale, et le mésoderme somatique la musculature segmentaire ainsi que le corps adipeux, le cœur et la gaine des gonades. Les cellules germinales primordiales sont représentées par 16 cellules polaires. Le système nerveux central dérive de neuroblastes ectodermiques ventro-latéraux et antéro-latéraux séparés de l’hypoderme; le système trachéal se forme à partir d’invaginations segmentaires paires de l’ectoderme. Le reste de l’ectoderme de l’embryon donne l’hypoderme; l’ectoderme extra-embryonnaire est résorbé lors de la fermeture dorsale. Il ne se forme ni amnios, ni séreuse.

4. La segmentation de l’œuf chez D. tryoni est caractéristique des Diptères. Le rythme des divisions synchrones et le mode de formation du blastoderme suivent des modalités constantes. Bien qu’on observe des variations dans le nombre des divisions, dans l’horaire de la colonisation de la surface de l’œuf par les noyaux, ainsi que dans le nombre et l’origine des cellules polaires et des vitellophages, néanmoins la structure du blastoderme édifié de même que la disposition des territoires présomptifs de ce blastoderme demeure vraisemblablement constant dans tout l’ordre. Les processus de gastrulation sont également constant, mise à part une différence dans le comportement des cellules polaires entre les Nématocères et les Cyclorhaphes. Au cours de l’organogenèse, il est de règle chez les Cyclorhaphes que les cellules polaires participent à la paroi du tube digestif, à l’opposé des Nématocères, en raison de la chronologie relative de la formation des cellules polaires et de la détermination des territoires présomptifs dans les 2 genres. L’édification du tube digestif chez D. tryoni est par ailleurs typique des Diptères, de même que l’évolution ultérieure du mésoderme. Les segments concernent surtout l’ectoderme et tendent à disparaître dans le mésoderme. La segmentation pré-orale est totalement supprimée. Les cellules germinales des gonades des Diptères dérivent toujours des cellules polaires, et le reste de la gonade du mésoderme. Chez les Nématocères, l’ectoderme extra-embryonnaire s’étend pour former un amnios et une séreuse; chez les Cyclorhaphes, il ne se développe pas de membranes embryonnaires. La destinée des diverses cellules dans la formation des organes larvaires des Diptères est plus ou moins constante d’une espèce à l’autre.

I would like to thank Dr. M. A. Bateman for the provision of material for this study, Miss S. McPhail for technical assistance, Mr. L. Congdon for assistance in the preparation of photomicrographs, and Dr. S. M. Manton and Professor D. R. Newth for advice on matters of presentation. The work was supported by a research grant from the University of Sydney.